Francesco Germano'
Automated Core Segmentation in Optical Images of Tumor Spheroids and Its Correlation with Drug Dose.
Rel. Gabriella Balestra, Samanta Rosati, Alessandra Introvaia. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Biomedica, 2026
Abstract
Every segmentation tool built to date for tumor spheroids or organoids targets only one boundary: where the object ends and the background begins. The border one layer inside, separating the core from the viable tissue around it, stays intact by that work, especially on ordinary bright-field images with no fluorescent channel to mark it. Aim of this study is, for the first time, to segment the core of glioblastoma multicellular tumor spheroids captured by a standard bench-top microscope, across two cellular compositions and four time points after treatment. Two segmentation methods were designed for the scope and scored against one shared, manually annotated ground truth, on a 32-image Test Set, kept separate from training and feature selection throughout.
The first is Region Growing (RG), implemented as the classical comparison baseline, driven by a seeded threshold with nothing learned on it
Relatori
Anno Accademico
Tipo di pubblicazione
Numero di pagine
Informazioni aggiuntive
Corso di laurea
Classe di laurea
URI
![]() |
Modifica (riservato agli operatori) |
